Deep Learning-Based Crack Detection in Concrete Bridge Decks and Pavements: A Experimental Review

Abstract

As transportation infrastructure gradually ages and maintenance costs continue to rise, efficient and precise defect detection of bridges, road surfaces, and concrete structures has become a critical factor in ensuring operational safety. Traditional manual crack identification methods are not only inefficient and subjective but also pose safety risks. The rapid advancement of computer vision and artificial intelligence technologies, particularly deep learning, offers effective solutions to address these challenges. This paper systematically reviews research on the application of deep learning models in typical tasks such as crack image classification, object localization, and pixel-level segmentation. It provides an in-depth analysis of the recognition mechanisms, technical advantages, and applicable scenarios of mainstream models, including convolutional neural networks and generative adversarial networks. Furthermore, it summarizes experimental findings and presents personal insights, aiming to provide reference and facilitate further research and application of deep learning in the intelligent maintenance and management of civil engineering infrastructure.